Neural network assisted cardiac auscultation

doi:10.1016/0933-3657(94)00026-O

Artificial Intelligence in Medicine

Volume 7, Issue 1, February 1995, Pages 53-66

https://doi.org/10.1016/0933-3657(94)00026-O Get rights and content

Abstract

Traditional cardiac auscultation involves a great deal of interpretive skill. Neural networks were trained as phonocardiographic classifiers to determine their viability in this rôle. All networks had three layers and were trained by backpropagation using only the heart sound amplitude envelope as input. The main aspect of the study was to determine what topologies, gain and momentum factors lead to efficient training for this application. Neural networks which are trained with heart sound classes of greater similarity were found to be less likely to converge to a solution. A prototype normal/abnormal classifier was also developed which provided excellent classification accuracy despite the sparse nature of the training data. Future directions for the development of a full-scale computer-assisted phonocardiographic classifier are also considered.

References (18)

L.E. Atlas et al.
Network learning modifications for multimodal classification problems: applications to EKG patterns
Neural Networks
(1988)
H.L. Baranek et al.
Automatic detection of sounds and murmurs in patients with Ionescu-Shiley aortic bioprostheses
Med. Bio. Eng. Comp.
(1989)
J.B. Barlow, Auscultation of the Heart. LKA 4421 (Decca/EMI,...
D. Barschdorff et al.
Heart sound analysis using neural and statistical classifiers: a comparison
1989 IEEE Computers in Cardiology
(1989)
H.E. Bassil et al.
Real time processing and analysis of fetal phonocardiographic signals
Clin. Phys. Physiol. Meas.
(1989)
J.R. Bulgrin et al.
Comparison of short-time fourier, wavelet and time-domain analyses of intracardiac sounds
Biomed. Sci. Inst.
(1993)
C.K. Friedberg
A.S. Gevins et al.
Applications of neural-network (NN) signal processing in brain research
IEEE Trans. Acoust. Speech Signal Process.
(1988)
G. Jamous et al.
Optimal time-window duration for computing time/frequency representations of normal phonocardiograms in dogs
Med. Bio. Eng. Comp.
(1992)

There are more references available in the full text version of this article.

Cited by (46)

An artificial intelligent-based model for detecting systolic pathological patterns of phonocardiogram based on time-growing neural network
2019, Applied Soft Computing Journal
Citation Excerpt :
Since the early 1990s when the “soft computing” officially became a study topic [1], many applied researches have been directed towards development of a reliable machine learning method to classify heart sound signal [2–7].
This paper presents a novel hybrid model for classifying time series of heart sound signal using time-growing neural network. The proposed hybrid model takes segmental behaviour of heart sound signal into account by combining two different deep learning methods, the Static and the Moving Time-Growing Neural Network, which we call STGNN and MTGNN, respectively. Flexibility of the model in learning both deterministic and stochastic segments of signal allows it to learn those complicated characteristics of heart sound signal caused by any obstruction on semilunar heart valve. The model is trained to distinguish between a patient group and a reference group. The patient group is comprised of the subjects with the semilunar heart valve abnormalities including aortic stenosis, pulmonary stenosis and bicuspid aortic valve, whereas the reference group which is composed of the individuals with the heart abnormalities other than those of the reference group or the healthy ones. The model is validated using two different databases: one comprised of 140 children with various heart conditions, and the other one constituted of 50 elderly patients with aortic stenosis. Both the datasets were collected from the referrals to the University hospitals. The overall accuracy and sensitivity of the model are estimated to be 84.2% and 82.8%, respectively. The results show that the model exhibits sufficient capability to distinguish between the patient and the reference group in such a demanding clinical application.
A novel heart-mobile interface for detection and classification of heart sounds
2018, Biomedical Signal Processing and Control
Citation Excerpt :
More recently, machine learning tools such as Support Vector Machine (SVM), Artificial Neural Network (ANN), and HMM have been used to classify heart sounds. ANN was one of the first widely used approaches to classify heart sounds [53]. Homomorphic filtering and k-means clustering are also used as feature extraction techniques with ANN to classify heart sounds into normal, systolic murmurs, and diastolic murmurs [55].
Diagnosis of heart disease requires that a medical practitioner investigate heart auscultations for irregular sounds, followed by echocardiography and electrocardiography tests. These expensive tests also require specialized technicians to operate. We present a low-cost, patient-centered device for the initial screening of the heart sounds that can be potentially used by the users on themselves. They can later share these readings with their healthcare providers. We have created an innovative mobile-health service platform for analyzing and classifying heart sounds.
The presented system enables remote patient-monitoring by integrating advanced wireless communications with a customized low-cost stethoscope. This system also permits remote management of a patient’s cardiac status while maximizing patient mobility. The smartphone application facilitates recording, processing, visualizing, listening to, and classification of heart sounds. We build our classification model using the Mel-Frequency Cepstral Coefficient and Hidden Markov Model. This application is tested in a hospital environment to collect live recordings from patients with positive results. The smartphone application correctly detected 92.68% of abnormal heart conditions in clinical trials at UT Southwestern Hospital.
Automatic diagnosis of septal defects based on tunable-Q wavelet transform of cardiac sound signals
2015, Expert Systems with Applications
Citation Excerpt :
The detection of septal defect can be based on classification between four classes: one includes the septal defects and others include normal, valvular defects, and other disorders. The classifiers which have been popularly and efficiently used for cardiac sound signals are neural network classifier (Cathers, 1995; Dokur & Ölmez, 2009; Gupta et al., 2007; Patidar & Pachori, 2015; Reed, Reed, & Fritzson, 2004) and SVM classifier (Ari, Hembram, & Saha, 2010; Choi & Jiang, 2010; Kao & Wei, 2011; Maglogiannis, Loukis, Zafiropoulos, & Stasis, 2009; Sun et al., 2014b). The TQWT is a recently developed wavelet transform for analysis of oscillatory signals (Selesnick, 2011).
Accurate and quick diagnosis of cardiac disorders like septal defects can be performed by automatic analysis of cardiac sound signals using advanced signal processing methods. In this paper, we present a new method for diagnosis of septal defects from cardiac sound signals using tunable-Q wavelet transform (TQWT) based features. To start with, the established constrained TQWT based approach has been used to derive the heart beat cycles from cardiac sound signals. Then the TQWT based decomposition of segmented heart beat cycles has been performed up to a certain level. The combinations of sub-bands obtained from TQWT based decomposition can be used to extract the diagnostic features. The correlation between sub-bands can characterize the various types of murmurs in cardiac sound signals. Therefore, in order to represent the murmurs in cardiac sound signals, proposed feature set was obtained with sum of average magnitude difference function (SAMDF) that have been computed from reconstruction of decomposed sub-bands. In search of effective feature set based on SAMDF that could provide significant classification performance, various decomposition levels have been examined. Moreover, in order to establish the usefulness of the proposed method for diagnosis of septal defects, besides cardiac sound signals for septal defects and normal, this study covers signals to be detected for valvular defects and other defects like ventricular hypertrophy, constrictive pericarditis etc. as available from publicly available datasets. The classification has been performed using least squares support vector machine (LS-SVM) with different kernel functions. At each decomposition level under study, the effect of quality-factor $(Q)$ of the TQWT from 1 to 50 on classification performance has been evaluated. Moreover, in order to show the effectiveness of the proposed method, results have been compared with existing method.
Detection of cardiac abnormality from PCG signal using LMS based least square SVM classifier
2010, Expert Systems with Applications
Auscultation, the technique of listening to heart sounds with a stethoscope can be used as a primary detection system for diagnosing heart valve disorders. Phonocardiogram, the digital recording of heart sounds is becoming increasingly popular as it is relatively inexpensive. In this paper, a technique to improve the performance of the Least Square Support Vector Machine (LSSVM) is proposed for classification of normal and abnormal heart sounds using wavelet based feature set. In the proposed technique, the Lagrange multiplier is modified based on Least Mean Square (LMS) algorithm, which in turn modifies the weight vector to reduce the classification error. The basic idea is to enlarge the separating boundary surface, such that the separability between the clusters is increased. The updated weight vector is used at the time of testing. The performance of the proposed systems is evaluated on 64 different recordings of heart sounds comprising of normal and five different pathological cases. It is found that the proposed technique classifies the heart sounds with higher recognition accuracy than competing techniques.
Support Vectors Machine-based identification of heart valve diseases using heart sounds
2009, Computer Methods and Programs in Biomedicine
Citation Excerpt :
The diagnostic classification of heart sound signals; since the present paper belongs to this heart sound research stream, we are going to analyse it in more detail. Part of these studies are dealing with the discrimination between normal (from healthy subjects) and abnormal (from subjects having a disease) heart sound signals [52–55], or with the discrimination between innocent and pathological murmurs in children [2,56–62]. Other studies are dealing with the detection from heart sound signals of particular heart diseases, such as coronary artery diseases [63–67] and heart valve diseases [68–75].
Taking into account that heart auscultation remains the dominant method for heart examination in the small health centers of the rural areas and generally in primary healthcare set-ups, the enhancement of this technique would aid significantly in the diagnosis of heart diseases. In this context, the present paper initially surveys the research that has been conducted concerning the exploitation of heart sound signals for automated and semi-automated detection of pathological heart conditions. Then it proposes an automated diagnosis system for the identification of heart valve diseases based on the Support Vector Machines (SVM) classification of heart sounds. This system performs a highly difficult diagnostic task (even for experienced physicians), much more difficult than the basic diagnosis of the existence or not of a heart valve disease (i.e. the classification of a heart sound as ‘healthy’ or ‘having a heart valve disease’): it identifies the particular heart valve disease. The system was applied in a representative global dataset of 198 heart sound signals, which come both from healthy medical cases and from cases suffering from the four most usual heart valve diseases: aortic stenosis (AS), aortic regurgitation (AR), mitral stenosis (MS) and mitral regurgitation (MR). Initially the heart sounds were successfully categorized using a SVM classifier as normal or disease-related and then the corresponding murmurs in the unhealthy cases were classified as systolic or diastolic. For the heart sounds diagnosed as having systolic murmur we used a SVM classifier for performing a more detailed classification of them as having aortic stenosis or mitral regurgitation. Similarly for the heart sounds diagnosed as having diastolic murmur we used a SVM classifier for classifying them as having aortic regurgitation or mitral stenosis. Alternative classifiers have been applied to the same data for comparison (i.e. back-propagation neural networks, k-nearest-neighbour and naïve Bayes classifiers), however their performance for the same diagnostic problems was lower than the SVM classifiers proposed in this work.
In search of an optimization technique for Artificial Neural Network to classify abnormal heart sounds
2009, Applied Soft Computing Journal
Artificial Neural Network (ANN) finds use in classification of heart sounds for its discriminative training ability and easy implementation. The selection of number of nodes for an ANN remains an important issue as an overparameterized ANN gets trained along with the redundant information that results in poor validation. Also a larger network means more computational expense, resulting more hardware and time related cost. Therefore, a compact and optimum design of neural network is needed towards real-time detection of pathological patterns, if any from heart sound signals. This work attempts to (i) design a compact form of output layer with less number of nodes than output classes, (ii) select a set of input features that are effective for identification of heart sound signals using Singular Value Decomposition (SVD), QR factorization with column pivoting (QRcp) and Fisher's F-ratio, (iii) make certain optimum selection of nodes in the hidden layer for a more effective ANN structure using SVD and (iv) select and prune weights based on the concept of local relative sensitivity index (LRSI) for empirically chosen overparameterized ANN structure for phonocardiogram (PCG) classification. It is observed that the proposed techniques perform better in terms of reduction of model residues and time complexity for classifying 12 different pathological cases and normal heart sound.

View all citing articles on Scopus

^∗: Fax (+61-2) 646 6520.

View full text

PaperNeural network assisted cardiac auscultation

Abstract

Network learning modifications for multimodal classification problems: applications to EKG patterns

Neural Networks

Automatic detection of sounds and murmurs in patients with Ionescu-Shiley aortic bioprostheses

Med. Bio. Eng. Comp.

Heart sound analysis using neural and statistical classifiers: a comparison

1989 IEEE Computers in Cardiology

Real time processing and analysis of fetal phonocardiographic signals

Clin. Phys. Physiol. Meas.

Comparison of short-time fourier, wavelet and time-domain analyses of intracardiac sounds

Biomed. Sci. Inst.

Applications of neural-network (NN) signal processing in brain research

IEEE Trans. Acoust. Speech Signal Process.

Optimal time-window duration for computing time/frequency representations of normal phonocardiograms in dogs

Med. Bio. Eng. Comp.

Paper
Neural network assisted cardiac auscultation